So we're united by our common loathing for The Other? Ugh. That's so bleak it's kind of funny.

I think these things are very closely related. It reminds me of the whole genre of "mixup" sitcoms, where one person thinks he's talking about X, and the other person thinks he's talking about Y, and hilarity ensues at the difference between those interpretations -- viewing X as Y and vice-versa.

References to certain local events (don't wanna dox myself) where the performers acted like the meaning of those events, and the context of the events, was X, when to me, the meaning is far away from X.

This sounds dumb in the abstract, I realize; but it's the kind of interpretive gulf that you're probably familiar with for anything that approaches politics. It's just that in this case, everything was so concrete, it's like being told a false story about something that happened in front of your eyes, and seeing other people agreeing with the story like it's consensus reality.

Appreciate the context.

Interesting to consider that having kids is, itself, one of the handful of anchors common to many people's experience -- something foundational and primal. The falloff in having children becomes, therefore, an additional cut line tethering us to each other.

Is there a way to know, with modest reliability, about when miners sell anything? Does anyone purport to?

It will be interesting to see what the reactions by artists -- and people who become artists -- wind up being. All art is a function of some inputs; when the inputs include AI art, and the context is a world full of AI art, what are the outputs?

hahaha

You've had an interesting afternoon.

Dust the 58k boys off, they're needed.

Do you generally think in terms of cycles, e.g., cycle high, then some pct of that for the low?

I find the halving-based cycles interesting, but only very coarsely useful. Much more interesting is trying to understand how the macro story has changed, and what the causal implications should be, but that's not usually on offer.

I hate it when they do that!

On another poll, presumably.

I appreciate the thoughtful analysis -- thanks for showing your thinking!

Ah. Well, I look fwd to the post.

A thing I find awe-inspiring is the idea (which I endorse) that, if all progress on training new frontier models stopped, we'd be exploiting the results of what we've already got for another decade. Makes my brain hurt to really think about it.

We'll know a lot more when the replication efforts bear fruit. There's some amount of accounting sleight-of-hand, but most sources I follow find the methodology credible, and you can run the thing on modest amounts of consumer hardware.

It has nothing to do w/ copyright, the issue is the method used to train the model. Details in the paper.

I actually mostly agree with this -- the points of contention might be how we each define "widely used" and "eventually". Without those parameter settings, the statement is basically un-falsifiable. But still I think he's saying something important, and why MoE can't be ignored.

I was wondering the same thing -- he doesn't even have to do anything, other people just need to interpret his signals in a manner that makes enforcement unlikely.

All that said, I agree that the CC situation was a great forcing function. If you couldn't manage to get non-custodial lightning set up, you probably just learned something about yourself, and we all (collectively) learned something about the ecosystem. Valuable lessons, both.

It will be interesting to see what the backlash will be in four years. If you pay attention, you will find the seeds being sown now, scattered in the dirt, waiting to sprout once they're forgotten.

Hot take, less than a week after the American President became an epic-level shitcoiner: bitcoiners will have severe nostalgia for the Gensler era before all is done.